To display a gray-scale image on a display panel such as a plasma display panel that is fundamentally only capable of two display states, a method is widely employed that separates one TV field period into sub-fields, assigns predetermined luminance weights to the sub-fields, and controls the presence or absense of light emission of each sub-field.
For instance, 256 levels of gray are represented by dividing a TV field period into eight sub-fields that are respectively given the luminance weights of “1”, “2”, “4”, “8”, “16”, “32”, “64”, and “128”. When an input image signal is an 8-bit digital signal, then the 8 bits are respectively assigned to the eight sub-fields starting with the least significant bit. Here, each sub-field image has two display states.
A CRT display bears the so-called inverse gamma property, so that while maximum luminance is proportional to “255”, minimum luminance is proportional to a decimal no greater than “1”. Hence the dynamic range is kept at a sufficient level of 255 or higher.
On the other hand, a plasma display panel has a linear luminous property, so that a gray level is expressed by a sum of luminance levels substantially proportional to sub-field weights. Which is to say, while maximum luminance is proportional to a sum of luminance weights of all sub-fields, i.e. “255”, minimum luminance is proportional to “1”. Because of this greater minimum luminance than that of the CRT, the dynamic range of the plasma display panel is narrower than the CRT.
The dynamic range of the plasma display panel may be widened by increasing the number of sub-fields so as to increase the number of reproducible levels of gray, but this technique is not easy to implement due to restrictuions such as discharge speeds of plasma display panels. Therefore, the number of sub-fields is normally limited.
Also, the aforementioned method of expressing 256 levels of gray using the eight sub-fields is known to be susceptible to halftone disturbances with significant false contours which appear when displaying a moving image.
To reduce such halftone disturbances, a technique has been devised that detects motion in an image and switches coding for each pixel or each image portion in the image.
As an example of this technique, coding is varied for each image portion such that when input is made in 256-level gray scale, light emission is effected in 256 levels of gray for a static image portion, while light emission is effected in a more limited number of gray levels for a moving image portion. In so doing, the moving image portion is coded so that the light-emission pattern changes with a certain degree of continuity against monotonous gray level changes of input image signals. This benefits a reduction of annoying false contours in the moving image display. Meanwhile, a desired sufficient gray scale is guaranteed in the static image display.
In such a conventional method, however, coding is switched at the boundary of the moving and static portions. In some images, this switching causes a certain impact on the boundary area. The impact of the switching is particularly well observed in boundaries of an object that is moving in plane within an image.